Fast Development of ASR in African Languages using Self Supervised Speech Representation Learning

Jama Hussein Mohamud; Lloyd Acquaye Thompson; Aissatou Ndoye; Laurent Besacier

Fast Development of ASR in African Languages using Self Supervised Speech Representation Learning

Sound 2021-03-17 v1 Computation and Language Audio and Speech Processing

Authors: Jama Hussein Mohamud , Lloyd Acquaye Thompson , Aissatou Ndoye , Laurent Besacier

Abstract

This paper describes the results of an informal collaboration launched during the African Master of Machine Intelligence (AMMI) in June 2020. After a series of lectures and labs on speech data collection using mobile applications and on self-supervised representation learning from speech, a small group of students and the lecturer continued working on automatic speech recognition (ASR) project for three languages: Wolof, Ga, and Somali. This paper describes how data was collected and ASR systems developed with a small amount (1h) of transcribed speech as training data. In these low resource conditions, pre-training a model on large amounts of raw speech was fundamental for the efficiency of ASR systems developed.

Keywords

automatic speech recognition speech recognition

Cite

@article{arxiv.2103.08993,
  title  = {Fast Development of ASR in African Languages using Self Supervised Speech Representation Learning},
  author = {Jama Hussein Mohamud and Lloyd Acquaye Thompson and Aissatou Ndoye and Laurent Besacier},
  journal= {arXiv preprint arXiv:2103.08993},
  year   = {2021}
}

Comments

Accepted at AfricaNLP2021 workshop at EACL 2021

Fast Development of ASR in African Languages using Self Supervised Speech Representation Learning

Abstract

Keywords

Cite

Comments

Related papers